Following Bertrand Russell, who interpreted the definite ...



John Justice

April 6, 2007

Singular-Term Semantics Simplified

1. Russell’s Division

Ever since Bertrand Russell made definite descriptions vanish into existential generalizations and contrasted them with simple symbols that directly designate individuals, many theorists have divided singular terms into two semantic types: genuine singular terms with individuals for semantic values and quantified noun phrases that Russell said had no “meaning in isolation.” Russell used the unrestricted quantifiers of standard predicate logic, which operate on open sentences to yield sentences. When sentences are translated using unrestricted quantifiers, there are no constituents identifiable as noun phrases.

Direct reference theorists have abandoned Russell’s eliminative analysis of quantified noun phrases, while embracing his separation of genuine singular terms from definite descriptions. Since generalized quantifiers make sets of sets the values of determiner-nominal phrases, the current view is that the values of definite descriptions are sets of sets, but the values of genuine singular terms are individuals.

I will argue that this division is a mistake. Singular terms have just one type of semantic value. They divide into names, deictic terms, and definite descriptions. They divide into rigid and nonrigid terms. But they do not divide into terms that have individuals for semantic values and terms that have sets for values.

Frege had not divided singular terms. He held that they all expressed sense conditions satisfiable by single objects. The objects satisfying these conditions were the terms’ referents, and they were also the terms’ semantic values. But Frege’s theory had two failings. First, he was unable to find nonsubjective sense conditions for names. Second, because the referent of a singular term was thought to be its semantic value, and a sentence’s truth-value depends on the values of its constituents, sentences with vacuous terms lacked truth-values.

Frege dismissed these problems as flaws in natural language, but neither can be dismissed by a satisfactory semantic theory. First, every term needs a publicly recognized way of determining its reference. Second, though speakers may presuppose that subjects of topic-comment sentences have referents, vacuous terms do not deprive sentences of truth-values. ‘I dined with the king of France’ is clearly false. To lack a referent is not to lack semantic value.

Russell’s division frees only definite descriptions from these difficulties, and it poses problems for the derivation of phrase values. Simple lexical terms and quantifier phrases are subjects of the same verb phrases and objects of the same verbs and prepositions. They occur together in coordinated noun phrases: ‘Tarzan and the ape’. The composition of semantic values in natural language does not seem to accommodate a division of terms.

However, some evidence appears to support the division. First, there is a structural difference between lexical noun phrases and noun phrases that have determiner-nominal syntax. Second, the lexical terms seem to lack the descriptive senses that would be needed if they were to share the semantics of complex noun phrases. Complex noun phrases characterize their referents, while simple terms seem to be directly referential. Third, the simple terms are rigid designators, and rigid designation might be linked to distinctive values.

I will show that these three observations, which seem to favor the Russellian division of singular terms, can be accounted for in a simple, unified semantic theory.

2. Noun-Phrase Semantics

Until Montague’s work on the semantics of noun phrases, there had seemed to be a sharp difference between the subject-predicate form of sentences with lexical subjects (“singular” sentences) and the quantified-formula form of sentences with determiner-nominal subjects (“general” sentences). However, Montague revealed that the dismissal of subject-predicate form had been too hasty.

In Montague Grammar, a quantified noun phrase has a value of its own: a set of possible verb-phrase extensions. For instance, the extension of ‘some boys’ is the set of all the possible verb-phrase extensions that contain at least one boy. ‘Some boys sing’ is true if the verb phrase’s extension belongs to the subject’s extension. In Montague semantics, noun-phrase extensions have elements containing individuals that do not appear in the extension of any noun-phrase constituent, and noun phrases are basically predicates applying to verb-phrase extensions. When a noun phrase occurs as the object of a verb or a preposition, the result is a mismatch of values that has to be repaired by changing the sentence’s syntax, by shifting the noun phrase’s extension, or by shifting the extension of the verb or preposition.

Mark Brown (1984) and Thomas McKay (1989) give quantified noun phrases values that are not sets of verb-phrase values and, consequently, do not require shifts in syntax or semantic value. In their account, a noun phrase’s extension comes out of its own nominal’s extension. It is the set of each subset of its nominal’s extension that meets the quantity condition set by the quantifier.

A determiner that is represented in logical notation by the universal quantifier makes a noun phrase that denotes[1] just its nominal’s extension:

(1) Val (all N) = { N }. [2]

If the nominal’s extension is empty, all N denotes the empty set. The empty set is its extension’s sole element.

The extension of some N is the set of all nonempty subsets of its nominal’s extension:

(2) Val (some N) = (N – { Ø }.

If the nominal’s extension is empty, some N denotes nothing. Its extension is the empty set.

A singular definite description denotes a singleton that contains the only member of its nominal’s extension:

(3) Val (the N) = {X: (X( = 1 & X=N}.

If the nominal’s extension is not a singleton, a description denotes nothing. Its extension is the empty set.

The sets selected by any quantified noun phrase for its extension are subsets of its restriction set, which is ordinarily its nominal’s extension. The nominal’s extension also works as a restriction set in composing the value of any phrase in which the noun phrase is a constituent.

In the composition of a sentence’s truth-value the domain of discourse is restricted to the subject nominal’s extension. Things outside of this set have no bearing on the sentence’s truth (Westerståhl 2001, 456). A sentence is true if and only if the subject denotes the intersection of its nominal’s extension and the verb phrase’s extension:

(4) (DN P) is true iff N ( P ( Val (DN).

A verb phrase constituted by a verb and its object selects individuals for its extension out of a restriction set supplied by the object noun phrase. For example, the extension of ‘sees no boy’ consists of individuals who are such that the boys whom they see is the set denoted by ‘no boy’.

(5) Val (V DN) = { x: N ( { y: ( Val (V) } ( Val (DN) }.

These evaluation rules cannot apply to terms that have individuals for extensions. If referents were the extensions of lexical terms, a second set of rules would be needed—and these rules would not assign values to vacuous terms or phrases containing them.

To get uniformity in noun-phrase values, Montague raised the value of a name from its referent to the set of verb-phrase extensions that contain its referent. But this Procrustean adjustment of value did nothing for terms that had no referent. What is needed is an account of noun-phrase meaning that gives every noun phrase, including vacuous terms, the same type of value—and does so without adjusting values.

3. Lexical-Term Values

Nominals and verb phrases have only sense conditions. They denote individuals of certain sorts, i.e. their extensions are sets of individuals. But noun phrases select the elements of their extensions by two constraints: a quantity condition and a sense condition. Noun phrases denote certain quantities of individuals of certain sorts. Their extensions contain right-sized sets of individuals of the right sort.

The key to a unified theory of noun-phrase semantics is uncovering the extensions of lexical terms. Lexical terms express both quantity and sense conditions. A name expresses the singular, reflexive condition of being its bearer. Some have doubted that bearing a name is a feature that distinguishes an individual because of an illusion that names are not really proper—do not have to belong exclusively to bearers. This illusion arises only if names are not recognized to be complex, abstract entities constituted of form, syntax, and reflexive meaning with origins in particular acts of naming. What can be shared are merely uninterpreted phonological or orthographical forms having no pertinent origins (Kaplan 1990, 110-11; Chierchia and McConnell-Ginet 2000, 77-78). The other lexical terms, the deictic terms, express relations to their own utterances: being the utterer, place, time, etc.

Names and singular deictic terms are singular definite noun phrases, differing from definite descriptions only in having simple forms and reflexive meanings. A singular noun phrase denotes only singletons. A definite noun phrase denotes only one nonempty set that contains every individual meeting its sense condition. So a singular, definite noun phrase denotes a singleton that contains the one individual that meets its sense condition, or else its extension is empty.

Names and deictic terms are suppletives: single, uninflected words that express both quantity and sense (Croft 2001, 356). A name expresses the unique reflexive condition of being its bearer.

A singular deictic term expresses some unique relation to its utterance. Because some relations to utterances are necessarily filled, some deictic terms never have an empty extension when evaluated at the circumstance of utterance. ‘I’, ‘here’, and ‘now’ are such terms.

A lexical term T denotes a singleton whose element is the sole member of the restriction set R determined by its sense condition:

(6) Val (T) = { X: (X( = 1 & X=R}.

If its restriction set is empty, a lexical term has the empty set as its extension.

The values of phrases with lexical-term subjects or objects are derived in the same way as the values of other phrases:

(4*) (T P) is true iff R ( P ( Val (T).

(5*) Val (V T) = { x: R ( { y: ( Val (V) } ( Val (T) }.

Since vacuous terms have empty extensions (denote no set), negative free logic is a consequence of this semantic analysis. Simple sentences with vacuous terms are false, and their negations are true. For example, ‘Santa brings gifts’ is false if ‘Santa’ has an empty extension.

4. Reference and Rigidity

A singular term refers if and only if it denotes, but its referent is not what it denotes. Denoting is the relation an expression has to the members of its extension. Nominals and verb phrases denote individuals, but they do not refer to them. Referring is the relation a definite noun phrase has to the individual, or individuals, in the single set that it denotes. To refer is to specify individuals. A definite noun phrase that denotes specifies each individual in the denoted set for attribution of distributive properties.

Singular reference has sometimes been thought to be the only sort of reference, but reference is not necessarily singular. Both Ns, the six Ns, and the Ns are definite noun phrases that denote sets having more than one member and that refer to each of those members in attributions of properties. To refer is to specify individuals—regardless of their numbers.

If a noun phrase is not definite, it may denote something other than a single, nonempty set. If it is universal, it denotes a single set—but it may be the empty set. If it is existential, like some Ns or two Ns, it denotes as many subsets of its restriction set as satisfy its quantity condition. Only definite noun phrases—those that denote one nonempty set—are referring terms.

A definite noun phrase refers rigidly if and only if it denotes the same set with respect to each circumstance at which it denotes a set. Proper names and deictic terms are rigid designators, but they are not the only rigid designators. Any nominal that cannot have a different (nonempty) extension when it is evaluated at a different circumstance makes a definite description refer rigidly. Nominals expressing essences of mathematical entities make rigid descriptions like ‘the least prime’, ‘both square roots of four’, and ‘the values of x2 + y2 = 1’. These nominals make what Kripke called strongly rigid designators. There are no circumstances at which they have empty extensions. Actualized descriptions and noun phrases with deictic determiners have utterance-reflexive sense conditions, which give them restriction sets that are unchanged or empty at each circumstance of evaluation.

The rigidity of lexical terms has the same source as the rigidity of other terms. Names and deictic terms are rigid because they express sense conditions that cannot be satisfied by different individuals at different circumstances. What individual bears a particular name, utters a particular utterance, or is the location of an utterance does not change when the circumstance of evaluation changes.

5. Conclusion

No singular term has an individual as its semantic value. Singular terms are definite noun phrases with the type of value that characterizes noun phrases. Lexical terms are unique in having reflexive meanings. These reflexive meanings give lexical terms two distinctive features. First, they do not need determiners since their quantity is entailed in the reflexive conditions. Second, they refer rigidly because their sense conditions are inalterable relations to linguistic entities.

Russell showed that sentences containing definite descriptions can be paraphrased as existential generalizations. But he was wrong in thinking, first, that the existential generalizations represent the true logical forms of these sentences, and second, that there are singular sentences that lack existential equivalents. The reflexive senses make existential paraphrases available for sentences containing lexical terms, but these paraphrases are merely logical equivalents with different syntax. They are not needed to distinguish either wide-scope sentence negation from narrow-scope term negation or de dicto attitude reports from de re reports. Scope distinctions can be marked as clearly in sentences containing definite descriptions or lexical terms as they can in sentences containing existential quantifiers.

Frege was right in thinking that all singular terms are of the same semantic type. But he did not recognize that lexical terms have reflexive meanings, that terms’ referents are not their semantic values, and that their values are of a type that is shared by all noun phrases.

Montague saw that all noun phrases need to have values of the same type; but the values he gave them—sets of verb-phrase extensions—incorporated extraneous elements, created type mismatches when noun phrases are objects, and failed to provide values for vacuous terms.

The view presented here this morning—that each noun phrase’s value is a set of subsets of its restriction set—heals Russell’s division of terms, supplies semantic values to all noun phrases, and simplifies the derivation of values for phrases in which noun phrases are constituents.

REFERENCES

Brown, Mark. (1984) “Generalized Quantifiers and the Square of Opposition,” Notre Dame Journal of Formal Logic 25, 303-322.

Chierchia, Gennaro and Sally McConnell-Ginet. (2000) Meaning and Grammar: An Introduction to Semantics, 2nd edition, MIT Press, Cambridge, MA.

Croft, William. (2001) “Typology,” in M. Aronoff and J. Rees-Miller (eds.), The Handbook of Linguistics, 337-368, Blackwell Publishers, Oxford.

Frege, Gottlob. (1892) “On Sinn and Bedeutung,” in M. Beaney (ed.), The Frege Reader, pp. 151-171, Blackwell Publishers, Oxford.

Kaplan, David. (1990) “Words,” The Aristotelian Society Supp. Vol. 64, 93-119.

McKay, Thomas. (1989) Modern Formal Logic, Macmillan Publishing Company, New York.

Montague, Richard. (1974) “The Proper Treatment of Quantification in Ordinary English,” in R. H. Thomason (ed.), Formal Philosophy: Selected Papers of Richard Montague, pp.247-270, Yale University Press, New Haven, CT.

Russell, Bertrand. (1905) “On Denoting,” Mind 14, 479-493.

Westerståhl, Dag. (2001) “Quantifiers,” in L. Goble (ed.), The Blackwell Guide to Philosophical Logic, pp. 437-460, Blackwell Publishers, Oxford.

-----------------------

[1] Expressions are said to denote the members of their extensions.

[2] Italic ‘D’, ‘N’, and ‘P’ are schematic letters for determiners, nominals, and verb phrases. Roman ‘N’ and ‘P’ designate extensions of nominals and verb phrases.

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download